Continued work on position unbiased ranking #5393
Conversation
chore: ignore swig jni file
run tests on pr to position_debias
add unbiased config params
feat: rough migration of acbull's source
chore: give better name to lambdarank bias regularizer
remove unused param position_bins_, replaced by truncation_level
shiyu1994
requested review from
StrikerRUS,
jameslamb,
guolinke and
jmoralez
as code owners
|
@shiyu1994 it's been 8 months since the last commit on this PR. Are you interested in pursuing it still? If you think it's valuable but don't have time yourself to work on it, could you find someone else at Microsoft who might be able to help? |
|
Updating my question from #5393 (comment). @shiyu1994 it's now been almost a year since the last commit on this PR. Are you interested in pursuing it still? If you think it's valuable but don't have time yourself to work on it, could you find someone else at Microsoft who might be able to help? And how does it relate to #5929 ? I'm trying to help but from the conversation in #5929, it feels to me like maybe you have a plan for learning-to-rank in LightGBM that's being discussed in private places and that I'm not aware of. |
|
@jameslamb This PR and #5929 are two different approaches to support unbiased ranking. The underlying algorithms are different. We think the method in #5929 is better. So, we prioritize #5929 first. I may come back to this when I have time, and change this PR into another option for unbiased ranking. |
|
Got it, ok thanks! Could we add a link to this PR in the list of feature requests (#2302) and close it for now? Since it's not actively being worked on. |
Sure. |
|
This pull request has been automatically locked since there has not been any recent activity since it was closed. To start a new related discussion, open a new issue at https://github.com/microsoft/LightGBM/issues including a reference to this. |
This is to replace #4531.